Niche Modeling: Predictions from Statistical Distributions

by Stockwell; David

• ISBN13: 9781584884941
• ISBN10: 1584884940
• Edition: 1st
• Format: Hardcover
• Copyright: 2006-12-15
• Publisher: Chapman & Hall/

Summary

Using theory, applications, and examples of inferences, Niche Modeling: Predictions from Statistical Distributions demonstrates how to conduct and evaluate niche modeling projects in any area of application. It features a series of theoretical and practical exercises for developing and evaluating niche models using the R statistics language. The author discusses applications of predictive modeling methods with reference to valid inferences from assumptions. He elucidates varied and simplified examples with rigor and completeness. Topics include geographic information systems, multivariate modeling, artificial intelligence methods, data handling, and information infrastructure.Above all, successful niche modeling requires a deep understanding of the process of creating and using probability. Off-the-shelf statistical packages are tailored exactly to applications but can hide problematic complexities. Recipe book implementations fail to educate users in the details, assumptions, and pitfalls of analysis, but may be able to adapt to the specific needs of each study. Examining the sources of errors such as autocorrelation, bias, long term persistence, nonlinearity, circularity, and fraud, this seminal reference provides an understanding of the limitations and potential pitfalls of prediction, emphasizing the importance of avoiding errors.

Table of Contents

0.1 Preface	xix
0.1.1 Summary of chapters	xix
1 Functions	1
1.1 Elements	1
1.1.1 Factor	1
1.1.2 Complex	2
1.1.3 Raw	2
1.1.4 Vectors	2
1.1.5 Lists	3
1.1.6 Data frames	3
1.1.7 Time series	3
1.1.8 Matrix	4
1.2 Operations	4
1.3 Functions	6
1.4 Ecological models	9
1.4.1 Preferences	11
1.4.2 Stochastic functions	11
1.4.3 Random fields	18
1.5 Summary	21
2 Data	23
2.1 Creating	24
2.2 Entering data	25
2.3 Queries	26
2.4 Joins	28
2.5 Loading and saving a database	29
2.6 Summary	29
3 Spatial	31
3.1 Data types	31
3.2 Operations	34
3.2.1 Rasterizing	37
3.2.2 Overlay	37
3.2.3 Proximity	39
3.2.4 Cropping	40
3.2.5 Palette swapping	40
3.3 Summary	44
4 Topology	45
4.1 Formalism	45
4.2 Topology	47
4.3 Hutchinsonian niche	47
4.3.1 Species space	48
4.3.2 Environmental space	48
4.3.3 Topological generalizations	49
4.3.4 Geographic space	49
4.3.5 Relationships	50
4.4 Environmental envelope	51
4.4.1 Relevant variables	51
4.4.2 Tails of the distribution	51
4.4.3 Independence	52
4.5 Probability distribution	52
4.5.1 Dynamics	53
4.5.2 Generalized linear models	54
4.6 Machine learning methods	57
4.7 Data mining	58
4.7.1 Decision trees	59
4.7.2 Clustering	59
4.7.3 Comparison	59
4.8 Post-Hutchinsonian niche	60
4.8.1 Product space	61
4.9 Summary	63
5 Environmental data collections	65
5.1 Datasets	66
5.1.1 Global ecosystems database	88
5.1.2 Worldclim	89
5.1.3 World ocean atlas	90
5.1.4 Continuous fields	90
5.1.5 Hydro1km	91
5.1.6 WhyWhere	91
5.2 Archives	91
5.2.1 Traffic	92
5.2.2 Management	92
5.2.3 Interaction	92
5.2.4 Updating	92
5.2.5 Legacy	92
5.2.6 Example: WhyWhere archive	93
5.2.7 Browsing	93
5.2.8 Format	94
5.2.9 Meta data	94
5.2.10 Operations	95
5.3 Summary	95
6 Examples	97
6.0.1 Model skill	97
6.0.2 Calculating accuracy	99
6.1 Predicting house prices	99
6.1.1 Analysis	100
6.1.2 P data and no mask	104
6.1.3 Presence and absence (PA) data	105
6.1.4 Interpretation	106
6.2 Brown Treesnake	107
6.2.1 Predictive model	107
6.3 Invasion of Zebra Mussel	109
6.4 Observations	113
7 Bias	115
7.1 Range shift	116
7.1.1 Example: climate change	116
7.2 Range-shift Model	117
7.3 Forms of bias	120
7.3.1 Width r and width error	120
7.3.2 Shift s and shift error	123
7.3.3 Proportional pe	123
7.4 Quantifying bias	123
7.5 Summary	125
8 Autocorrelation	127
8.1 Types	128
8.1.1 Independent identically distributed (IID)	128
8.1.2 Moving average models (MA)	128
8.1.3 Autoregressive models (AR)	129
8.1.4 Self-similar series (SSS)	129
8.2 Characteristics	130
8.2.1 Autocorrelation Function (ACF)	130
8.2.2 The problems of autocorrelation	136
8.3 Example: Testing statistical skill	137
8.4 Within range	139
8.4.1 Beyond range	139
8.5 Generalization to 2D	140
8.6 Summary	141
9 Non-linearity	143
9.1 Growth niches	144
9.1.1 Linear	145
9.1.2 Sigmoidal	145
9.1.3 Quadratic	147
9.1.4 Cubic	154
9.2 Summary	155
10 Long term persistence	157
10.1 Detecting LTP	159
10.1.1 Hurst Exponent	162
10.1.2 Partial ACF	163
10.2 Implications of LTP	166
10.3 Discussion	171
11 Circularity	173
11.1 Climate prediction	173
11.1.1 Experiments	174
11.2 Lessons for niche modeling	177
12 Fraud	179
12.1 Methods	181
12.1.1 Random numbers	181
12.1.2 CRU	184
12.1.3 Tree rings	186
12.1.4 Tidal Gauge	186
12.1.5 Tidal gauge - hand recorded	188
12.2 Summary	190
References	191
Index	199